intel/auto-round

Pull Requests Commits

Support HunyuanImage model quantization

lvliang-intel committed 76 days ago

1c29a5cd

skip quantizing mtp.fc since vLLM doesn't support (#1731)

xin3he committed 77 days ago

Verified b3ef561d

Revert "ignore mtp.fc for qwen3_5 due to vllm failure (#1710)" (#1730)

xin3he committed 77 days ago

Verified c2344554

Support new model Qwen/Qwen3.6-35B-A3B (#1705)

lvliang-intel committed 78 days ago

Verified 2dda035b

refine int4 doc (#1720)

wenhuach21 committed 78 days ago

Verified 5b6f7ea7

[research feature] Introduce INT4 support at the algorithm level (#1641)

wenhuach21 committed 78 days ago

Verified 11e91869

ignore mtp.fc for qwen3_5 due to vllm failure (#1710)

xin3he committed 78 days ago

Verified 8428b984

add support for gemma4 model (#1655)

n1ck-guo committed 79 days ago

Verified 55977e7b

Update gaudi-docker to v1.24.0 & fix CUDA UT (#1708)

XuehaoSun committed 79 days ago

Verified 8d784505

update mtp quant for special cases (#1691)

xin3he committed 80 days ago

Verified 89769e8a

rename scheme INT8_W8A8 to INT8 (#1687)

thuang6 committed 81 days ago

Verified 6a3bb90f

fix: add missing run_mllm entry point alias (#1695)

JGSphaela committed 83 days ago

Verified 693e4bd7

Add Claude skills for AutoRound (#1686)

lvliang-intel committed 83 days ago

Verified 4abe4067

Enable low_cpu_mem_usage for mxfp/nvfp (#1648)

Kaihui-intel committed 83 days ago

Verified 8d7bb84c

Gate large FP4 packing test by GPU memory (#1696)

yiliu30 committed 83 days ago

Verified a4dee986

support WOQ model input, such as kimi2.5 (#1642)

xin3he committed 84 days ago

Verified d5a0097d

Support diffusion model AIDC-AI/Ovis-Image-7B quantization (#1616)

lvliang-intel committed 84 days ago

Verified 95a84d2b

fp8_block bug fix (#1693)

mengniwang95 committed 84 days ago

Verified 42101801

reduce flux CUDA CI tuning memory from 30G to 5GB (#1694)

xin3he committed 84 days ago

Verified 751216a4

Fix hadamard transform weight dtype, using float32 as default and in-place transformed weight. (#1665)

lkk12014402 committed 84 days ago

Verified 43748dd9

Stabilize Qwen3-Omni MoE weight fidelity test for matching NaNs (#1681)

lvliang-intel committed 85 days ago

Verified 865f31b2

Enable NextStepDiffusion and support multi-device tuning for diffusion (#1640)

xin3he committed 86 days ago

Verified f81e20d6

Support MXINT4 scheme (#1666)

mengniwang95 committed 87 days ago

Verified c817d495

[mllm] support longcat_next (#1637)

xin3he committed 87 days ago

Verified 0821817b

Add compressed-tensors format export support for W4A16 and W8A16 (#1669)

thuang6 committed 87 days ago

Verified bbda38a6

Refactor module access to use PyTorch get_submodule / set_submodule (#1590)

scopophobic committed 90 days ago

Verified 58221b4a

Fix omni model test CI issue (#1667)

lvliang-intel committed 90 days ago

Verified bbda4046

Update unit test requirements (#1670)

XuehaoSun committed 91 days ago

Verified 440faaf1

Fix missing extra_config export for unsupported ignore_layers like `mlp.gate` (#1660)

lvliang-intel committed 91 days ago

Verified d2a62b89

import gptqmodel in confest to avoid error in transformers (#1668)

xin3he committed 92 days ago

Verified 106a3223

Older